US20090006858A1 - Secure seed provisioning - Google Patents

Abstract

A method is used for secure seed provisioning. Data is derived from inherent randomness in an authentication device. Based on the data, the authentication device is provisioned with a seed.

Description

BACKGROUND OF THE INVENTION
• Computer networks, and in particular Wide Area Networks (WANs) such as the Internet, provide opportunities for the misuse and abuse of communications traveling thereover. For example, two users (e.g., a human user and an enterprise server) communicating via the WAN may have their communications intercepted and/or altered. Also, it is possible for one user to misrepresent his, her, or its identity to another user.
• Thus, there is a need for both privacy and authentication between users of the network communicating with one another. In other words, users should be able to rely on the fact that their transmissions will not be intercepted or altered, and that transmissions from someone purporting to be a particular user do in fact originate from that user.
• In many secure communication applications, a seed is required in order to perform certain cryptographic operations such as encryption, decryption, authentication, etc. The seed may comprise, by way of example, a symmetric key or other secret shared by two or more entities.
• One such application is in authentication tokens, such as the RSA SecurID® authentication token commercially available from RSA Security Inc. of Bedford, Mass., U.S.A. The RSA SecurID® authentication token is used to provide two-factor authentication. Authorized users are issued individually-registered tokens that generate single-use token codes, which change based on a time code algorithm. For example, a different token code may be generated every 60 seconds. In a given two-factor authentication session, the user is required to enter a personal identification number (PIN) plus the current token code from his or her authentication token. This information is supplied to an authentication entity. The authentication entity may be a server or other processing device equipped with RSA ACE/Server® software, available from RSA Security Inc. The PIN and current token code may be transmitted to the authentication entity via an encryption agent equipped with RSA ACE/Agent® software, also available from RSA Security Inc. If the PIN and current token code are determined to be valid, the user is granted access appropriate to his or her authorization level. Thus, the token codes are like temporary passwords that cannot be guessed by an attacker, with other than a negligible probability.
• A given RSA SecurID® token typically contains one or more seeds that are utilized in computing the token outputs. The authentication entity performing the verification of the token outputs requires access to one or more seeds associated with the token in question. Typically, such authentication entities have access to the same seed or set of seeds that the token uses to generate its output.
SUMMARY OF THE INVENTION
• A method is used for secure seed provisioning. Data is derived from inherent randomness in an authentication device. Based on the data, the authentication device is provisioned with a seed.
• One or more implementations of the invention may provide one or more of the following advantages.
• Seed provisioning of an authentication token or other device may be performed off site with sufficient security, control, and accountability, and, at least in some cases, without requiring a unique identification number to be built in.
• These and other features and advantages of the present invention will become more readily apparent from the accompanying drawings and the following detailed description.
BRIEF DESCRIPTION OF THE DRAWINGS
• FIG. 1 is a block diagram of an example system for use with secure seed provisioning.
• FIG. 2 is a flow diagram of a procedure for use with secure seed provisioning.
DETAILED DESCRIPTION
• Described below is a technique for use in the configuration and seeding of security tokens (e.g., RSA SecurID® authentication tokens) at third party facilities, particularly at facilities of a configuration agent, such that a token can be configured without the configuration agent having security-defeating knowledge about the token. In particular, it is desirable to allow third parties to provision a token with a seed, but in such a way that the third party will not know, or be able to construct, the seed after the seed provisioning process is complete. As noted above, the seed may comprise, by way of example, a symmetric key or other secret shared by two or more entities.
• In addition, it is desirable to make it difficult or impossible for the third party to create clones of a token. Two or more tokens are considered cloned if they generate the same series of pseudo-random numbers in the same order.
• For example, the third party may be a contract manufacturer assembling tokens. Use of the technique described herein allows the contract manufacturer (after assembling the token) to provision the seed to the token such that the contract manufacturer does not know, and does not need to be responsible for, the resulting seed value of the token.
• FIG. 1 illustrates an example of asystem100 for use with the technique. It is to be appreciated, however, thatsystem100 is merely an example implementation, and the invention is not restricted to use in this or any other particular system configuration.
• FIG. 1 illustratessystem100 in which a robotic and/ormanual programming station105 communicates with a token (device)110 and a central control (non third party)station115 acrossrespective communication links185,180. (The particular number of stations and tokens shown is by way of example only, and a given system in which the invention is implemented may include more or fewer than the particular number shown.) Each of thecommunication links180,185 may be or include one or more of the following: a global computer network such as the Internet, a wide area network (WAN), a local area network (LAN), a satellite network, a telephone or cable network, a wireless link such as 802.11 and/or Bluetooth, a USB based link, a serial or parallel link, a processor interface link, a memory interface link, or various portions or combinations of these and other types of links. In at least some implementations, one or more oflinks180,185 may rely on a secure connection established therein, in a conventional manner, through the SSL protocol.
• In at least one specific implementation, there is a difference betweenlinks185,180:link185 connects the programming station to the token, and may be a relatively simple link such as a proprietary bus, or a chip interface such as IIC or SPI;link180 is between the programming station and the central control station, and may be use the Internet, a dialup link, or a physical media transfer such as a CD-ROM or tape.
• Station105 may be at and/or controlled by a configuration agent, andstation115 may be at and/or controlled by RSA Security Inc. or other central control. The programming station is responsible for helping to make the token ready for use, in particular by helping to program a seed into the token, and for communicating with the central control station as described below. The programming station may include and be controlled by acomputer system160 having amemory165.
• In the example implementation,token110 includes aprocessor120 having anoscillator section125 driven by acrystal circuit130 that includes acrystal135 andcapacitors140. The processor has or is connected to read only memory (ROM)145 containing firmware instructions for the processor, and has or is connected to read-write memory (RAM)150. The processor is powered by a battery155 (in other implementations, power may be supplied in addition or instead by another power source such as a USB port). Depending on the implementation as described below, the token may or may not have a counter170 driven by the oscillator section, and/or a unique identification number such as processor'sunique identifier175. (In another example implementation, e.g., for an event-synchronous token, a simple RC (resistor-capacitor) driven oscillator may be adequate—and in some cases the oscillator is entirely internal to the microprocessor. All such examples still have variations in processor instruction timing due to tolerances and therefore can be used with the technique.)
• With respect to the technique, it is assumed that a potential attacker can monitor the communications which occurs between the token and the programming station. In addition, it is assumed that the potential attacker is able to examine the contents of the memory in the computer system, if any, controlling the station.
• It is assumed that firmware in the token has not been reverse engineered, and that security at the central control station's production environment has not been compromised.
• In a first type of the technique, it is assumed that the token has no unique identification number available such as a processor identifier. Thus in particular it is assumed that every token has an identical processor which has no individual unique identifier. In at least some implementations each processor is identical in that the firmware instructions contained in the ROM (or other memory) of the processor are identical from unit to unit, and there are no unique identifiers stored in the memory of the processor. (It is also to be noted that in at least implementations some or all of the firmware instructions may be contained in other types or configurations of memory such as EEPROM, FLASH memory, or RAM, instead of or in addition to the ROM.)
• Theoretically in at least some cases, if two tokens have processors that are identical, and configuration data is identical, the two tokens configured with the same data could be expected to generate identical pseudo random number streams.
• In at least some implementations, the technique may rely on some uniqueness for each token based on at least one of the following sources of variability (i.e., inherent randomness):
• 1. The processor relies on the oscillator section which runs at a frequency which is determined by a number of electronic component based parameters including the characteristics and tolerances of the oscillator section's crystal, the characteristics and tolerances of the capacitors, and tolerances in the characteristics of internal gates of circuitry of the oscillator section. There may also be some other variables including the stray capacitance of a printed circuit board or other substrate on which token components are disposed.
• 2. There is non uniformity among tokens in powerup characteristics, e.g., the amount of time (latency) between when the token's battery is connected to the processor (e.g., is soldered onto a printed circuit board of the token), which essentially starts the oscillator section running, and when the token is configured.
• 3. The token's RAM reliably may power up with random data.
• 4. The token may have a built in random number generator.
• In at least some implementations, the technique uses at least the first two sources of variability. In at least one example implementation, counter170 is or includes a large counter, e.g., 32 bits wide, which is started as soon as the battery is connected to the processor (e.g., by soldering). This counter is incremented by a software (e.g., firmware instruction) loop having an execution speed that is controlled by the frequency of the oscillator section. The component-based parameters for each token cause the oscillator section, and therefore the loop, to operate at a rate which varies slightly from token to token.
• When the token receives a configuration command, which is not synchronized with battery connection, the counter is stopped and sampled, which yields a counter value (which may be used as R_internaldiscussed below). The combination of the slightly different oscillator frequency and the non uniformity in latency between battery connection and token configuration as described above yield counter values which vary with significant unpredictability from token to token.
• In other implementations other ways may be used instead or as well to generate or help generate a unique counter value per token.
• The unique counter value per token may be used to help provide the following features:
• • Two tokens programmed with the same configuration data end up with different seeds.
  • The third party performing the configuration of the token, who potentially can read all the data going into and coming out of the token, cannot determine the resulting seed from this data.
  • Central control can re-construct the seed so that a token seed record can be created.
  • The resulting seed maintains a high entropy.
• In the example implementation, the first type of the technique can be used to help provide these features, by:
• • Using the counter R_internal, discussed above,
  • Using a key K_class, known by the token and by central control, but not known to the third party, and
  • Injecting a high quality random number R_externalduring configuration.
• In the example implementation:
• • R_internalis n-bits (e.g., 32 bits) of pseudo-random data generated by the device (e.g., in the form of the counter value described above, or based on the contents of RAM at powerup),
  • R_externalis 128-bits of random data loaded from an external source of randomness,
  • K_classis 128-bit secret that is common to all tokens within a given class (e.g., a token type produced by a specific manufacturer),
  • N is the serial number assigned to the token,
  • O is an output of the token which can be used by central control to reconstruct the token's seed (e.g., master seed), and
  • S is the token's seed.
• With respect to the example implementation, the technique may by executed as follows.
• 1. The token internally generates Rinternai.
• 2. R_externalis generated by the configuration agent.
• 3. R_externalis sent to the token from the configuration agent.
• 4. The configuration agent assigns serial number N to the token.
• 5. The token computes K_class′=K_classXOR R_internal(left aligned) XOR N (right aligned). (The data blocks for the values of R_internaland N should be aligned such that the R_internaldata lines up with padding from N and vice versa, e.g., 2D E7 FF FF FF FF FF FF XOR FF FF FF FF FF FF 12 34.) (“XOR” denotes bit-wise addition modulo two.)
• 6. The token encrypts R_externalunder key K_class′ to produce O
• 7. The token outputs O to the configuration agent
• 8. The token encrypts O under K_classto produce seed S
• 9. The token stores seed S
• 10. The configuration agent sends pairs of serial numbers N and outputs O to central control.
• 11. Central control computes S from O and K_class
• In at least some implementations, the following characteristics are provided and are important:
• • R_internal, K_class, K_class′, and S are all kept secret. (In at least some cases the technique assumes an attacker can see R_external, O, and N, but this is not enough for the attacker to compute S.)
  • If R_internalis implemented as a counter value, it is not reset to zero when the token is reset. A specific example implementation operates as follows. The counter is stopped when the token receives a configure command, and the current counter value is used as R_internal. The counter remains stopped, but retains its current value until the token receives a reset command. Upon receiving a reset command, the token once again starts the counter, but starts counting from where it left off. When the next configure command comes in, the counter is once again stopped to read R_internal, but again the current value is kept in case the token is reset at some time in the future. Alternatively, it is possible to keep the counter running all the time, but in at least some implementations this arrangement would be impractical as it would quickly run down the battery.
  • The configuration agent knows only O and R_external.
  • O is unique to every device.
  • After configuration, the token retains the R_internalvalue used by the technique. When the token is issued a reset command to prepare it to be re-configured, the token starts counting from where it left off, and then stop counting when a new configuration command is received. This helps prevent an attack in which the attacker issues a software reset command with the purpose of resetting R_internalso that it can be stopped in a known state by the rapid issuance of a new configuration command after the software reset command.
• In a second type of the technique, it is assumed that the token has a unique identification number available such as a processor identifier. The unique identification number may only be read internally by the processor that contains K_classand processor instructions implementing the technique. The unique identification number makes each token fundamentally unique.
• In at least one implementation, each processor starts off with a chip serial number as well as a unique value U_internalprogrammed in at the time of chip fabrication by the semiconductor manufacturer. It is possible to use this unique value as the seed if there is a case in which it is deemed an acceptable risk that the processor manufacturer knows the mapping between the processor's chip serial number and the unique value. In cases in which using the unique value as the seed is unacceptably risky, an additional step involving a value generated at the third party location is needed. It is possible to use an implementation which does not require K_class, but including K_classadds additional protection against a potential risk that the third party would collude with the chip manufacturer, and also makes the second type of the technique more consistent with the first type described above.
• In the second type of the technique, the token uses unique value U_internalinstead of R_internalto produce seed S, the semiconductor manufacturer provides central control with the mapping between the chip serial number and the unique value, and the third party provides central control with the mapping between the token serial number and R_external. Thus, only central control and the token have all the elements needed to produce seed S.
• In at least some implementations, the following characteristics are provided and are important:
• • N, U_internal, R_external, and K_classare required for the token to compute the seed S.
  • Pairs of chip serial numbers and U_internalare known only to the semiconductor manufacturer and central control.
  • Pairs of token serial numbers and R_externalare known only to the configuration agent and central control.
  • The semiconductor manufacturer knows only chip serial numbers and U_internal.
  • The configuration agent knows only U_internaland R_external.
• In at least some implementations, the entropy source for the generation of R_externalmust be unpredictable and hardware based, a minimum of 128 bits of entropy are required for each value of R_external, and the random number generator (RNG) must have passed the Diehard tests for randomness as referenced in Marsaglia, G., 1995, Diehard Battery of Tests of Randomness, http://stat.fsu.edu/geo/diehard.html. See also Rukhin, A., Soto, J., Nechvatal, J., et al, 2001, A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications, NIST Special Publication 800-22, http://csrc.nist.gov/publications/nistpubs/800-22/sp-800-22-051501.pdf.
• Although reference is made herein to random numbers, it is to be appreciated that the invention can be implemented using pseudorandom numbers or other types of strings of sufficient entropy.
• The term device or computer as used herein refers generally to any processor-based equipment or system capable of implementing at least a portion of the technique as described herein.
• One or more oftoken110 andstations105,115 may be, be included in, or include, by way of example and without limitation, a computer, a mobile telephone, a personal digital assistant (PDA), a smart card, an authentication token, a server, and/or various portions or combinations of these and other processing devices. One or more oftoken110 andstations105,115 may thus be implemented as otherwise conventional processing devices programmed by software and/or firmware to perform portions of the technique as described herein. Conventional aspects of such equipment are well known to those skilled in the art, and are therefore not described in detail herein.
• In an example implementation, the token comprises or is otherwise associated with an authentication token, such as an RSA SecurID® authentication token. However, the technique is adaptable in a straightforward manner to a wide variety of other cryptographic processing devices.
• Station105 may communicate withtoken110 and/orstation115 directly overrespective links185 and/or180, or may communicate via one or more intermediary processing devices. For example, if the token comprises an authentication token, it may communicate withstation105 overlink185 using an intermediary device such a desktop or portable personal computer, mobile telephone or PDA. Such intermediary devices are not specifically shown inFIG. 1, but token110 may be viewed as comprising, for example, a combination of an authentication token and an associated computer or other intermediary device. Similarly, one or more ofstations105,115 may incorporate or communicate through one or more intermediary processing devices. As indicated above, the term “processing device” as used herein is intended to encompass such combinations of devices.
• Details regarding certain conventional cryptographic techniques suitable for use in conjunction with the present invention may be found in, e.g., A. J. Menezes et al., Handbook of Applied Cryptography, CRC Press, 1997, which is incorporated by reference herein.
• It should again be emphasized that the technique implementations described above are provided by way of illustration, and should not be construed as limiting the present invention to any specific embodiment or group of embodiments. For example, the invention can be implemented in other types of systems, using different arrangements of processing devices and processing operations. Also, message formats and communication protocols utilized may be varied in alternative embodiments. Moreover, various simplifying assumptions made above in the course of describing the illustrative embodiments should also be viewed as exemplary rather than as requirements or limitations of the invention. Numerous alternative embodiments within the scope of the appended claims will be readily apparent to those skilled in the art.

Claims (21)

1. A method for use in secure seed provisioning, the method comprising the steps of:

deriving data from inherent randomness in an authentication device; and

based on the data, provisioning the authentication device with a seed.

2. The method ofclaim 1, wherein the seed comprises a symmetric key.

3. The method ofclaim 1, wherein the inherent randomness is based on a powerup characteristic of the authentication device.

4. The method ofclaim 1, wherein the inherent randomness is based on an electronic component characteristic of the authentication device.

5. The method ofclaim 1, wherein the inherent randomness is based on an amount of time between unsynchronized events in the authentication device.

6. The method ofclaim 1, further comprising:

starting to increment a counter when power is applied to the authentication device,

wherein the data is based on a value of the counter.

7. The method ofclaim 1, further comprising:

further deriving the data from a source of randomness external to the authentication device.

8. The method ofclaim 1, further comprising:

outputting an intermediate level value to a recipient, the intermediate level value being required for the recipient to calculate the seed.

9. The method ofclaim 1, further comprising:

deriving a value from the inherent randomness; and

outputting the value to a recipient, the value being required for the recipient to calculate the seed.

10. The method ofclaim 1, further comprising:

outputting a first value to a first recipient, the first value alone being insufficient to allow calculation of the seed; and

at the first recipient, calculating the seed from the first value and a second value already available to the first recipient, the second value being unavailable to a second recipient.

11. The method ofclaim 1, further comprising:

deriving the seed from a unique value provided in the authentication device.

12. The method ofclaim 1, further comprising:

using the inherent randomness to affect a value of a counter in the authentication device.

13. The method ofclaim 1, further comprising:

avoiding resetting the value of a counter in the authentication device to zero when the authentication device is reset.

14. The method ofclaim 1, further comprising:

deriving the seed from a source of randomness that is external to the authentication device, the source of randomness being unpredictable and hardware based.

15. The method ofclaim 1, further comprising:

deriving the seed using encryption.

16. The method ofclaim 1, further comprising:

deriving the seed from a unique value programmed into the authentication device at the time of chip fabrication.

17. The method ofclaim 1, further comprising:

controlling access to values such that only one or more authorized entities outside the authentication device have sufficient information to calculate the seed.

18. The method ofclaim 1, further comprising:

deriving the seed based on a mapping between a chip serial number and a unique value programmed into the authentication device at the time of chip fabrication.

19. The method ofclaim 1, further comprising:

deriving the seed based on a mapping between a serial number of the authentication device and a value based on an external source of randomness.

20. A method for use in secure seed provisioning, the method comprising the steps of:

deriving data from a secret number embedded in multiple authentication devices including a first authentication device, and from a unique number embedded in the first authentication device only; and

based on the data, provisioning the first authentication device with a seed.

21. A method for use in secure seed provisioning, the method comprising the steps of:

deriving a first number from an internal source of randomness;

receiving a second number based on an external source of randomness;

receiving a serial number;

internally retrieving a non-unique secret key;

deriving a third number from the first number, serial number, and the non-unique secret key;

externally exposing the third number; and

deriving a key from the second and third numbers.

Images